A Rust-native port of Hugging Face Transformers providing ready-to-use NLP pipelines and transformer models like BERT, GPT2, and T5.
rust-bert is a Rust library that provides native implementations of state-of-the-art Natural Language Processing models and pipelines. It ports Hugging Face's Transformers library to Rust, enabling tasks like translation, summarization, text generation, and question answering with high performance and multi-threading support. It solves the problem of integrating advanced NLP capabilities into Rust applications without relying on Python bindings.
Rust developers and engineers building NLP-powered applications, such as chatbots, content analysis tools, or language translation services, who need performance and native integration.
Developers choose rust-bert for its native Rust performance, GPU acceleration, and comprehensive model support, offering a seamless alternative to Python-based NLP libraries without sacrificing capabilities.
Rust native ready-to-use NLP pipelines and transformer-based models (BERT, DistilBERT, GPT2,...)
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Offers pre-built pipelines for tasks like QA and translation, as shown in README examples that work in just a few lines of code, reducing boilerplate.
Compatible with popular architectures like BERT, GPT-2, and T5, with a detailed matrix in the README linking models to tasks such as summarization and NER.
Supports GPU inference via LibTorch bindings, enabling faster processing for large models, noted in benchmarks for text generation tasks.
Requires manual installation of libtorch with environment variable configuration, which is error-prone and adds deployment friction, as detailed in the README's platform-specific steps.
As a port of Hugging Face's library, it may lag behind the latest model releases and fine-tunes, limiting access to state-of-the-art advancements.
Relies on large downloads like libtorch (several GBs for CUDA) and models, increasing resource requirements and build times, especially for embedded systems.